private void cboRules_SelectedIndexChanged(object sender, EventArgs e)
{
btnRequest.Enabled = true;
lblMsg.Text = "";
txtRepContent.Text = "";
if (string.IsNullOrEmpty(this.cboRules.SelectedValue.ToString()))
{
txtTitleRule.Text = "";
txtContentRule.Text = "";
txtListRule.Text = "";
txtRequestUrl.Text = "";
isOk = false;
return;
}
isOk = true;
txtRequestUrl.Text = this.cboRules.SelectedValue.ToString();
var rule = Spider.GetRuleByName(this.cboRules.Text);
txtTitleRule.Text = rule.TitleRule;
txtContentRule.Text = rule.ContentRule;
txtListRule.Text = rule.ListRule;
encoding = rule.Encoding;
}
private void CatchJokesAction(Action <string> action)
{
int endPage = 1;
int startPage = 1;
if (cbIsRepeat.Checked)
{
startPage = txtStart.Text.ToInt32();
endPage = txtEndPage.Text.ToInt32();
for (int i = startPage; i <= endPage; i++)
{
action(string.Format("正在抓取:{0}", string.Format(txtRequestUrl.Text, i)));
List <JokeInfo> jokes = new List <JokeInfo>();
var content = Spider.GetHtmlContent(string.Format(txtRequestUrl.Text, i), encoding);
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(content);
string jokelistXPath = txtListRule.Text.Trim();
var nodes = doc.DocumentNode.SelectNodes(jokelistXPath);
if (nodes == null || nodes.Count == 0)
{
lblMsg.Text = "没有抓取到集合数据，请重新定义规则";
continue;
}
else
{
lblMsg.Text = nodes.Count.ToString();
txtRepContent.Text = nodes[0].InnerHtml;
//return;
}
JokeInfo jokeinfo;
HtmlNode temp;
foreach (var node in nodes)
{
temp = HtmlNode.CreateNode(node.OuterHtml);
jokeinfo = new JokeInfo();
jokeinfo.Title = temp.SelectSingleNode(txtTitleRule.Text.Trim()).InnerText.Trim();
jokeinfo.Content = temp.SelectSingleNode(txtContentRule.Text.Trim()).InnerText.Trim();
if (string.IsNullOrEmpty(jokeinfo.Content) || string.IsNullOrEmpty(jokeinfo.Title))
{
continue;
}
jokes.Add(jokeinfo);
}
if (jokes.Count == 0)
{
lblMsg.Text = "没有抓取到数据，请重新定义规则!";
btnRequest.Enabled = true;
return;
}
jokeLogic.AddJokes(ToJokes(jokes));
lblMsg.Text = "抓取数据成功";
}
}
else
{
action(string.Format("正在抓取:{0}", txtRequestUrl.Text));
List <JokeInfo> jokes = new List <JokeInfo>();
var content = Spider.GetHtmlContent(txtRequestUrl.Text, encoding);
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(content);
string jokelistXPath = txtListRule.Text.Trim();
var nodes = doc.DocumentNode.SelectNodes(jokelistXPath);
if (nodes == null || nodes.Count == 0)
{
lblMsg.Text = "没有抓取到集合数据，请重新定义规则";
btnRequest.Enabled = true;
return;
}
else
{
lblMsg.Text = nodes.Count.ToString();
txtRepContent.Text = nodes[0].InnerHtml;
//return;
}
JokeInfo jokeinfo;
HtmlNode temp;
foreach (var node in nodes)
{
temp = HtmlNode.CreateNode(node.OuterHtml);
jokeinfo = new JokeInfo();
jokeinfo.Title = temp.SelectSingleNode(txtTitleRule.Text.Trim()).InnerText.Trim();
jokeinfo.Content = temp.SelectSingleNode(txtContentRule.Text.Trim()).InnerText.Trim();
if (string.IsNullOrEmpty(jokeinfo.Content) || string.IsNullOrEmpty(jokeinfo.Title))
{
continue;
}
jokes.Add(jokeinfo);
}
if (jokes.Count == 0)
{
lblMsg.Text = "没有抓取到数据，请重新定义规则!";
btnRequest.Enabled = true;
return;
}
jokeLogic.AddJokes(ToJokes(jokes));
lblMsg.Text = "抓取数据成功";
}
btnRequest.Enabled = true;
lblCatchStatus.Text = "";
}